1

Thesis Abstract to obtain the Master of Science Degree in Biological Engineering

June 2016

CO2 Separation using Polymeric Ionic Liquids Membranes: the effect of

mixing different Cyano Anions

Raquel Marinho Teodoro dos Santos1*

1 Instituto Superior Técnico, Lisboa, Portugal

* Corresponding Author: raquel.marinho.teodoro.santos@tecnico.ulisboa.pt

Article Info

Abstract

Keywords:

CO2 Separation

Ionic Liquids

Polymeric Ionic Liquids

Composites

Cyano-based Anions

Gas Permeation

In order to reduce CO2 emissions and further prevent air pollution and global warming,

sustainable and efficient CO2 separation processes must be developed. Membrane-based

technologies represent a simple and environmental friendly alternative to traditional CO2

separation methods. Having received a large amount of attention in the last years, polymeric

ionic liquid (PIL) membranes combine the benefits of membrane technology with the unique

proprieties of ionic liquids (ILs). In this work several PIL-IL composite membranes were

prepared. Both PIL and IL have either cyano-functionalized ([N(CN)2]-, [C(CN)3]- or

[B(CN)4]-) or [NTf2]- counter anions, with the anions of the PIL and IL being different from

one another. The four synthesized PILs used in this work have a pyrrolidinium polycation

backbone, while the five ILs have either an imidazolium ([C2mim]+) or a pyrrolidinium

([Pyr14]+) based cation. Several experimental conditions were tested in order to achieve the

maximum number of homogenous and free standing PIL-IL composite membranes. The CO2

and N2 separation proprieties (permeability, diffusivity and solubility) of suitable membranes

were evaluated at a fixed temperature (293 K) and constant trans-membrane pressure

differential (100 kPa) using a time-lag method so that trends regarding the different counter

anions could be evaluated. From all 42 PIL-IL composite membranes prepared, 21 were

suitable for gas permeation experiments and 3 surpassed the 2008 Robeson upper bound for

CO2/N2 separation performance. These high performance membranes contain the [C(CN)3]-

and [B(CN)4]- counter anions, enlightening therefore the promise these anions entail for

future high CO2 separation performances membranes.

1. Introduction

Since the industrial revolution until the present

time, there has been an increased need for electricity

generation and consumption. Despite the existence of

cleaner alternative energy sources, fossil fuels are still

the world’s primary energy source and are expected to

remain so for the next couple of years. One concern of

the burning of these fuels is the emission of

anthropogenic carbon dioxide (CO2) which, in turn, is

largely responsible for air pollution and global

warming. [1] One of the solutions proposed to reduce

the emission of CO2 to the atmosphere consist of

Carbon Capture and Storage (CCS) systems, which

can be defined as a set of technologies that allow the

capture of CO2 emitted from the burning of fossil fuels

in power plants (like coal and natural gas), as well as

from other industrial processes such as cement, iron

and steel manufacture. After the separation of the CO2

at these point sources, the gas is pressurized in order

to be transported to a storage site. [2, 3] Although CO2

transportation and storage present some technological

and economic challenges, it is the capture of CO2 that

still needs an additional effort in research and

development so that alternative economic, energetic

and environmental viable methods can be

implemented. [4] Therefore, the efficient separation of

CO2 from other gases, namely methane (CH4),

nitrogen (N2) and hydrogen (H2), represents a major

technical, economic and environmental challenge.

Since power plants are unquestionably the major

source of anthropogenic CO2, the majority of the

studies addresses power plant CO2 streams conditions.

Within a power plant, four main strategies for CO2

capture have been proposed, depending on the

different point sources: pre-combustion (CO2 is

removed before combustion takes place), post-

combustion (CO2 is removed after combustion of

fossil fuels), oxy-fuel combustion (the fuel is

combusted in the presence of nearly pure oxygen,

resulting in a stream with high concentration of CO2)

and chemical looping combustion (direct contact

between fuel and air is avoided through the use of a

solid oxygen carrier). [2, 3] Among all the

technologies used so far for CO2 separation, four stand

out as the most relevant: solvent chemical absorption,

psychical/chemical adsorption, cryogenic distillation

and membranes. Because of its simplicity, membranes

2

(which can be porous or dense) have many benefits

including reduced environmental impact, low

operation costs and energy requirements, small scale

of equipment and easiness of integration into already

existing processes. [5] The majority of commercially

available membranes are organic membranes, also

known as polymeric membranes, and their

applications include post-combustion (CO2/N2), pre-

combustion (CO2/H2) and biogas (CO2/CH4)

separation. [6, 7] ILs offer a novel platform for CO2

separation technologies and in 2001 they were first

proposed as alternative solvents for CO2 separation by

Blanchard. [39]

Ionic Liquids (ILs) are salts comprised by organic

cations and inorganic or organic anions that possess a

set of unique chemical and physical properties which

make them alternative solvents already used in

numerous applications. [11] These properties include

low melting point, low volatility and flammability,

high thermal stability and electric conductivity.

However, the most important feature of ILs is perhaps

their tunability, which allows for the design of ILs that

can be tailored to fit in a particular technology

application. [11, 12]

The combination of membrane technology with

ILs results in different membrane configurations and

morphologies, namely polymeric ionic liquid (PIL)

membranes. [13] PILs belong to a subclass of

polyelectrolytes in which the polymeric backbone is

formed by monomer repeating units, each one

containing one ionic liquid specie, that overall form a

macromolecular architecture. [14] PILs combine the

advantages of both polymers and ionic liquids and thus

they have also been finding applications in different

technological fields, including analytical chemistry,

[15] biotechnology, [16] materials science, [17] and

gas separation, [18] among others. Neat PIL

membranes represent the simplest configuration of

PIL based membranes. Tomé and coworkers

concluded that neat PILs with a pyrrolidinium-based

backbone did not outperform the imidazolium

analogues. Despite this, the CO2 separation

performance stayed in the same order of magnitude.

[19] PIL membranes are usually mechanically and

thermally stable but their gas permeation proprieties

are hindered. [12] In order to overcome this

disadvantage, PIL-IL composite membranes represent

an alternative to neat PIL membranes. A PIL-IL

composite membrane contains a PIL framework in

which a certain percentage of free IL is incorporated.

Ionic liquids bearing cyano-functionalized anions are

reported to have low viscosity values and membranes

containing these anions are reported to have high CO2

separation performances, on top or above the 2008

Robeson upper bound. [27]

In 2015, Tomé and coworkers, synthesized

pyrrolidinium-based neat PIL membranes, with three

different cyano counter anions: dicyanamide

([N(CN)2]-), tricyanomethane ([C(CN)3]-) and

tetracyanoborate ([B(CN)4]-). All the three membranes

obtained were very brittle and broke easily.

Afterwards, PIL-IL composite membranes were

prepared by incorporating different percentages of free

IL (20 wt%, 40 wt% and 60 wt%) into the prepared

PILs. The PIL-IL composite membranes containing

the [C(CN)3]- counter anion, for the incorporation of

all the three different percentages of free IL resulted in

stable and homogenous membranes. By increasing the

amount of free IL, the CO2 permeability increased and

for 60 wt% of free IL the 2008 Robeson upper bound

for CO2/N2 pair was surpassed. [20] Giving the

outstanding results obtained by Tomé et al., cyano-

functionalized counter anions seem to hold a great

promise to obtain stable and over performing CO2

separation membranes.

The work carried out in this thesis is a

continuation of their work where the novelty here is

the preparation of membranes with different counter

anions in the PIL (Figure 1) and the IL (Figure 2). In

parallel, PIL-IL composite membranes with the

bis(trifluoromethylsulfonyl)imide ([NTf2]-) as a

counter anion, in both the PIL and IL, were also

synthesized and their CO2 separation performance

evaluated using the same procedures as those used for

the cyano-based membranes.

2. Experimental Section

2.1 Materials

Poly(diallydimethyammonium) chloride solution

(average Mw 400,000-500,000, 20 wt% in water) was

supplied by Sigma-Aldrich. The salts sodium

dicyanamide (NaN(CN)2, >97 wt%), sodium

tricyanomethane (NaC(CN)3, 98 wt%), and lithium

bis(trifluoromethylsulfonyl)imide (LiNTf2, 99%), as

well as the ionic liquids 1-ethyl-3-methylimidazolium

dicyanamide ([C2mim][N(CN)2], >98 wt%), 1-ethyl-

3-methylimidazolium tricyanomethane

([C2mim][C(CN)3], >98 wt%), 1-ethyl-3-

methylimidazolium bis(trifluoromethylsulfonyl)imide

([C2mim][NTf2], 99 wt%) and 1-butyl-1-

methylpyrrolidinium

bis(trifluoromethylsulfonyl)imide ([Pyr14][NTf2], 99

wt%) were purchased from IoLiTec GmbH. Merck

KGaA (Germany) provided the 1-ethyl-3-

methylimidazolium tetracyanoborate

([C2mim][B(CN)4], >98 wt%), while the potassium

tetracyanoborate (KB(CN)4) was synthesized as

reported elsewhere. [21] Air Liquide supplied the

carbon dioxide (CO2) and nitrogen (N2) with, at least,

99.99% purity. These gases were used with no further

purification.

3

Figure 1 - Chemical structure of the four PILs used in this work.

2.2 PILs Synthesis

The four PILs with a pyrrolidinium-based

polycation and cyano functionalized or [NTf2]-

counter anions (Figure 1), were obtained by anion

exchange reactions (Figure 3) from the commercially

available polyelectrolyte precursor,

poly(diallyldimethylammonium chloride), followed

by further purification steps. [20] In a typical

procedure, for instance to obtain approximately 15 g

of PIL C(CN)3, an aqueous solution of

poly(diallyldimethylammonium chloride) (55 g, 0.068

mol, 150 mL of distilled water) was added and mixed

to an aqueous solution of NaC(CN)3 (8.5 g, 0.071 mol,

20 mL in distilled water) in a round bottom flask. The

final solution was stirred for, at least, 30 minutes at

room temperature. A 5 wt% excess of NaC(CN)3 salt

was added to the solution to ensure equimolar anion

exchange (Figure 3). Due to its hydrophobic nature,

PIL C(CN)3 precipitated as a white solid (Figure 1) in

the aqueous media, just as soon as it was formed. The

polymer was then washed with distilled water, filtered

(Sartorius Stedim, Germany) and dried at 45 ºC in an

oven with forced air convection (VENTI-Line), until

constant weight was obtained. The other two

polymers, PIL B(CN)4 and PIL NTf2, were obtained

using the same procedure, but with the respective salts

KB(CN)4 and LiNTf2, as illustrated in Figure 3. The

hydrophilic polymer PIL N(CN)2 did not precipitate in

water and thus required a different purification

method. After removing the water by rotary

evaporation (VWR, IKA RV 10) at 45 ºC, PIL N(CN)2

and NaCl, a byproduct of the anion exchange reaction,

are left behind in the solid state. In order to dissolve

the polymer and precipitate the NaCl, 250mL of

ethanol were added. The excess of precipitated salt

was filtered and the filtrate was kept at -5 ºC overnight

to complete the NaCl precipitation. The remaining

NaCl in suspension was filtrated in the next day. This

procedure was repeated until no more NaCl precipitate

was observed. After evaporation of ethanol, a dark

yellow solid corresponding to PIL N(CN)2 was finally

obtained (Figure 1).

Figure 2 - Chemical structure of the five ILs used in this work.

Figure 3 - Anion exchange reaction used to synthesize the four different PILs studied in this work.

4

2.3 Membrane Preparation Method

The preparation of PIL-IL composite membranes

was attempted by a solvent casting method combining

the synthesized PILs (Figure 1) with commercial ILs

(Figure 2). Solutions containing the PIL and the

corresponding amount of free IL were prepared using

appropriated solvents and afterwards, the prepared

solutions were stirred until both PIL and IL were

completely dissolved and a homogeneous solution was

obtained. The PIL and IL solutions were then poured

into dishes and left for slow evaporation of the

solvents (Figure 4). Finally, to ensure that the solvent

was completely evaporated, the membranes were

dried at 45 ºC before gas permeation measurements. In

order to obtain stable and homogenous membranes,

different solvent casting conditions were tested

including the use of different solvents (acetone,

acetonitrile, ethanol, dimethyl sulfoxide (DMSO) and

dimethylformamide (DMF)), PIL and IL (w/v)%

concentrations, evaporation times and temperatures

(Figure 4), as well as different plate materials

(polytetrafluoroethylene (PTFE) and glass). The initial

casting conditions were taken from Tomé et al.

previous study [20] and as a starting point, the

membranes were prepared in either acetone or

acetonitrile, using 6 (w/v)% PIL and IL solutions, and

left to evaporate in Petri dishes between 2 to 4 days at

room temperature.

Figure 4 - Solvent evaporation (a and b) at room temperature and

(c) above room temperature during the casting procedure used to

prepare PIL-IL composite membranes.

2.4 Gas Permeation Experiments

The single gas (CO2 and N2) permeation

experiments were carried out, at 293 K, with a trans-

membrane pressure differential of 100 kPa using the

time-lag method. A time-lag apparatus (Figure 5)

allows for the simultaneous determination of

permeability and diffusivity while gas solubility

values were determined using Eq. 2. The apparatus

(positioned inside a thermostatic cabinet, where

temperature is controlled with a precision of ±0.05 K)

is composed by two stainless steel tanks, the feed tank

(5 dm3) and the permeate tank (34.2 ± 0.2) cm3, both

connected to a flat-type permeation cell with an

effective area of 13.9 cm2.

In a typical procedure, vacuum is first applied to

the whole system (at least for 12h) to ensure that

possible traces of water and gases are removed from

the feed and permeate side, as well from the membrane

itself. Second, the vacuum (<0.1 kPa) is isolated on the

permeate side, to ensure the initial gas concentration

on this side is approximately null (C0≈0), and on the

feed side the desired gas (CO2 or N2) is introduced

until the desired feed pressure (100 kPa) is achieved.

Finally, the single gas permeation experiments were

conducted. At least three separate experiments of each

gas on a single membrane sample were carried out.

Between each run, the permeation cell and lines were

evacuated, on both upstream and downstream sides

until the pressure was below 0.1 kPa. The thicknesses

of the PIL-IL composite membranes measured in this

work (120 to 200 µm) were measured before and after

testing using a digital micrometer (Mitutoyo, model

MDE-25PJ, Japan). Average thickness was calculated

from six measurements taken at different locations of

each membrane sample. At the end of the gas

experiments, no residual IL was found inside the

permeation cell and similarly, the membranes mass

remained constant.

Figure 5 - Schematic representation of the time-lag apparatus. P

represents the pressure sensors, V the manual valves, VF the feed

tank, VP the permeate tank and T a thermostatic air bath. [19]

Dense membranes separate gases through the

solution-diffusion mechanism. Gas permeability

(Eq.1) of a pure gas passing through a dense

membrane is defined as the thickness normalized

steady-state gas flux (𝐽) under a transmembrane

pressure difference (∆𝑝 = 𝑝1 − 𝑝2):

𝑃 = 𝐽𝑙

∆𝑝 (1)

where 𝑙 is the membrane thickness and 𝑝2 and 𝑝1 are

the upstream and downstream pressure, respectively.

5

Gas permeability (Eq.2) reflects the ability of a

gas to permeate the membrane. According to the

solution-diffusion model, gas permeability is the

product of gas diffusivity (D) and solubility (S) across

that membrane:

𝑃 = 𝑆 × 𝐷 (2)

Diffusivity is usually expressed in cm2.s-1 and it

accounts for the gas ability to move through the

membrane material. Several properties of the

membrane material are directly related with gas

diffusivity including the polymer free volume and

chain flexibility. Solubility is often expressed in cm3

(STP).cm-3.cmHg-1.

3. Results and Discussion

3.1 Membrane Forming Ability

Table 1 summarizes all the membranes prepared

in this work as well as their stability and homogeneity.

Figure 6 depicts the 21 stable and homogenous

membranes obtained in this work (membranes marked

with Ѵ on Table 1).

Overall, it was difficult to obtain stable and

homogenous composite membranes using PIL

N(CN)2, essentially due to its hydrophilic nature,

being the membranes with free IL C(CN)3 an

exception (Figure 6 – a to c) due to its excellent

compatibility with the [N(CN)2]- anion.

Regarding membranes bearing PIL C(CN)3, the

[C(CN)3]- and [B(CN)4]- anions differ only in a cyano

group, which makes them more similar in terms of

volume, allowing the formation of stable and

homogenous membranes with free IL B(CN)4 (Figure

6 – d to f). The [C(CN)3]- and [N(CN)2]- anions also

differ in a cyano group, which also makes them

similar, resulting in two suitable membranes (Figure 6

– g and h), but the hydrophilic nature of the [N(CN)2]-

anion compromises the mechanical stability of the PIL

C(CN)3 – 60 IL N(CN)2 membrane (Table 1). Despite

the chemical and geometrical differences between

[NTf2]- and [C(CN)3]- anions, the [NTf2]- anion (either

from the IL[Pyr14][NTf2] or IL([C2mim][NTf2]) is

highly flexible and thus allow for the dispersion

between PIL C(CN)3 chains and can prompt the

formation of homogeneous membranes up to 40 wt%

of free IL incorporated (Figure 6 – i to l). This

behavior is consistent with the well-known high

conformational structural flexibility of the [NTf2]-

anion. [22, 23, 24]

Composite membranes containing PIL B(CN)4

with free IL N(CN)2 are all heterogeneous (Table 1)

due to the rigid geometry of the [B(CN)4]- anion and

its incompatibility with the [N(CN)2]- anion. In

general, if a PIL is able to incorporate 60 wt% of a

certain free IL, this PIL should also be able to

incorporate 40 and 20 wt% of the same free IL.

However, the membrane PIL B(CN)4 – 20 IL C(CN)3

is heterogeneous (Table 1) while its 40 and 60 wt%

analogues are homogenous and stable (Figure 6 – m

and n). The membrane PIL B(CN)4 – 20 IL B(CN)4

was reported by Tomé et al. to be heterogeneous and

brittle [20] and, as stated before, the [C(CN)3]- and

[B(CN)4]- anions are very similar. Therefore, the

chemical similarity between the membranes PIL

B(CN)4 – 20 IL C(CN)3 and PIL B(CN)4 – 20 IL

B(CN)4 could explain the segregation of the latter

membrane. It seems that for composites of PIL B(CN)4

with higher amounts (40 and 60 wt%) of free IL

C(CN)3, the presence of higher amount of [C(CN)3]-

anions increases the compatibility of the starting

materials and thus homogeneous membranes can be

obtained. Membranes with PIL B(CN)4 with free IL

IL[C2mim][NTf2] (Figure 6 – o and p) present the

same behavior than its analogues bearing PIL C(CN)3.

The first thing to be noted for the PIL NTf2

composite membranes is that, for all the three studied

ILs having cyano-functionalized counter anions only

20 wt% of these free ILs can be incorporated into PIL

NTf2 (Figure 6 – s to u) so that mechanically stable and

homogenous membranes can be obtained. A possible

explanation might be that [NTf2]- anions can be more

strongly connected to the PIL polycation (due to the

localized charge of the PIL polycation), somehow

allowing for less polymer chain mobility and thus for

a more packed macromolecular structure with little

space for entrapping the free IL’s cations and anions,

allowing for, in general, only 20 wt% of the three free

ILs containing cyano-functionalized anions to be

incorporated within the polymeric chains of PIL NTf2.

Table 1 - Summary of all the membranes prepared in this work. Cells with a gray background represent membranes already reported in previous

works. [19, 20] N.S stands for “not synthesized”. Membranes denoted with a Ѵ mark are stable and homogenous, while those with the mark

X are non-stable and/or heterogeneous membranes.

6

Figure 6 - Pictures of the prepared PIL-IL composite membranes that, after the solvent casting process, are stable and homogenous. Circles:

red – PIL N(CN)2 composite membranes; green – PIL C(CN)3 composite membranes; orange – PIL NTf2 composite membranes.

Although Tomé et al. reported the incorporation of 60

wt% of the free IL [Pyr14][NTf2] into PIL NTf2 and

obtained a stable and homogenous membrane, [19] in

the present work only up to 40 wt% of the free IL

[C2mim][NTf2] could be incorporated inside the same

PIL (Figure 6 – q and r). Both PIL NTf2 cation and the

IL [Pyr14][NTf2] cation are based on pyrrolidinium

moieties, while the IL [C2mim][NTf2] cation is based

on imidazolium moieties. These difference in cations

is the key factor regarding these membranes forming

ability.

From all 42 membranes prepared 21 are stable

and homogenous while the other 21 are either

heterogeneous, non-stable or both. From the 21

prepared composite membranes containing the

[C(CN)3]- anion (either as PIL C(CN)3 or free IL

C(CN)3)), 15 (71 %) of them formed free standing and

homogeneous. By applying the same statistics to the

other anions, a 45 % success rate is achieved for the

composite membranes containing [NTf2]- anion, 44 %

for membranes bearing [B(CN)4]- anion and only 33%

for the composite membranes combining [N(CN)2]-

anions. These simple statistic calculations show that

PIL-IL composites containing [C(CN)3]- anions are

the most successful in the preparation of stable and

homogenous composite membranes, due to the better

structural and/or chemical compatibility of the

[C(CN)3]- anion with the other cyano-functionalized

anions, and also with the [NTf2]- anion, as well as its

geometrical flexibility. At the other end, membranes

bearing the [N(CN)2]- anion are the most difficult to

prepare. In general, the high hydrophilic nature of this

anion result in phase separated and/or gel-like

membranes, impossible to be manipulated, bringing

the success rate to the lowest (33 %) amongst the four

anions studied in this work.

3.2 Gas permeability, diffusivity and solubility

All 21 membranes from Figure 6 were measured

in the time-lag apparatus (Figure 5) and their CO2 and

N2 permeabilities, selectivities and diffusivities

obtained. Two trends were observed: first, the CO2

permeability values are always higher than those of N2

permeability for all the composite membranes, thanks

to the difference between CO2 and N2 solubilities,

being the CO2 solubility values higher than the N2

solubility values. This difference justify the use of

these membranes in CO2/N2 separation due to their

selective separation behavior. Second, by increasing

the amount of free IL incorporated into the composite

membranes, both the CO2 and N2 permeabilities

increase. This increment in gas permeability can be

attributed to the increase of both gas solubility and

diffusivity, being the increase of gas diffusivity way

more significant. The larger contribution of diffusivity

to the increase in the permeability values indicates an

increase in free volume of these PIL-IL composite

membranes, which enhances the polymer chain

mobility, thanks to the presence of the free ions pairs.

Both these trends have been observed for other PIL-IL

composite membranes [20, 25]. Adding to these two

trends, composite membranes with the same PIL

containing either free IL C(CN)3 or IL B(CN)4 always

outperformed composite membranes containing one

of the two ILs bearing an [NTf2]- counter anion. This

trend was already reported for other membranes based

on these ILs [26, 28] and the difference in the ILs

viscosity (being the ILs bearing an [NTf2]- counter

anion more viscous), as well as the high

conformational structural flexibility of the [NTf2]-

anion, which results in more packed membranes, are

presented as possible reasons for this trend.

7

Figure 7 - Gas permeabilities (top; 1 Barrer = 10-10 cm3(STP)cm cm-2 s-1 cmHg-1), diffusivities (middle) and solubilities (bottom) through

several composite membranes. The data regarding the membrane PIL C(CN)3 – 60 IL C(CN)3 was taken from Tomé et al. [20]. Error bars

represent standard deviations based on three experimental replicas.

8

Composite membranes bearing either PIL

C(CN)3 or PIL NTf2 [19] with the IL [Pyr14][NTf2]

always outperformed their analogues with IL

[C2mim][NTf2], as it was reported before for other

membranes [28]. The high viscosity of the IL

[Pyr14][NTf2] (twice that of the IL [C2mim][NTf2]) and

the localized charge of the pyrrolidinium based cation

are possible explanations for the diverse CO2

separation performance.

From all the composite membranes studied in this

work, three can be considered to have high CO2

separation performances, more precisely: PIL N(CN)2

– 60 IL C(CN)3, PIL C(CN)3 – 60 IL B(CN)4 and PIL

B(CN)4 – 60 IL C(CN)3. Adding to these three is the

PIL C(CN)3 – 60 IL C(CN)3 membrane previously

studied [20]. The presence of free IL C(CN)3 in three

of the four membranes shows the great versatility and

compatibility of this IL with other ions. The [C(CN)3]-

anion is also present in the PIL of two membranes,

meaning that the presence of this anion, either in the

free IL or PIL, leads to high CO2 separation

performance. The CO2 diffusivity through PIL C(CN)3

– 60 IL C(CN)3 is slightly higher than that of the other

membranes (Figure 7, middle), while the two

membranes with the [B(CN)4]- anion present slightly

higher CO2 solubility values (Figure 7, bottom). By

changing either the PIL C(CN)3 or the IL C(CN)3 of

the membrane PIL C(CN)3 – 60 IL C(CN)3 by PIL

B(CN)4 or IL B(CN)4, respectively, an increase in CO2

permeability is always achieved (Figure 7, top),

mainly due to an increase in the CO2 solubility, which

in turn compensates the decrease in CO2 diffusivity.

This result shows that the source (PIL or IL) of the

[B(CN)4]- is not so relevant here since the CO2

permeability, diffusivity and solubility values of the

two membranes having this anion are very similar. The

increase in CO2 solubility for membranes containing

the [B(CN)4]- anion, which resulted in higher CO2

permeabilities, was already reported. [25] The strong

CO2 molecules ̶ [B(CN)4]- anion interactions, as well

as the weak cation-anion interactions, may be the main

key factor for increased CO2 solubilities. [25] Also, it

has also been observed that the increasing number of

cyano groups of the IL’s anion leads to higher CO2

permeabilities. [26]

3.3 CO2 Separation Performance

In order to better evaluate the CO2 separation

performance of these composite membranes, a

Robeson Plot is used. The membranes with the highest

CO2 separation performances fall above or on top the

upper bound, on the upper-right corner of the plot.

The ideal selectivity (Eq.3), also known as

permselectivity or permeability selectivity (αi/j), is a

measure of how well a membrane material discerns

one gas from another. Like permeability, it is a

property of the membrane material and it can be

determined by dividing the permeability of most

permeable gas i (𝑃𝑖) by the permeability of the least

permeable gas j (𝑃𝑗):

𝛼𝑖𝑗⁄=𝑃𝑖𝑃𝑗= (

𝐷𝑖𝐷𝑗) × (

𝑆𝑖𝑆𝑗) (3)

Membranes with better gas separation

performance have higher permeability values for a

single gas specie within a gas mixture and a higher

permselectivity value as well. Although this is the

desired scenario, it has been proven that a trade-off

relationship exists between these two parameters,

meaning, membranes that are more selective are

usually less permeable and vice versa. [8] This trade-

off was described in 1991 by Robeson. By plotting the

permeability of the most permeable gas against the

permeability selectivity, on a log-log scale, the

existence of an upper bound was shown. [9] Since

1991 many new membranes were obtained and tested,

and their respective data released, and so the upper

bound was revised by Robeson in 2008 for numerous

gas pairs (CO2/N2, CO2/CH4, O2/N2, etc.). [10] Eq.4

describes the upper bound:

𝑃𝑖 = 𝑘𝛼𝑖𝑗⁄

𝑛 (4)

where 𝑃𝑖 is the permeability of the most permeable gas

in the gas mixture, αi/j represents the gas pair

permselectivity and n is the upper bound slope. The

upper bound represents an empirical correlation based

on the compilation of extensive experimental results

for several membranes.

Table 2 - CO2 permeability (1 Barrer = 10-10 cm3(STP)cm cm-2 s-

1 cmHg-1) and permselectivity values obtained for the four PIL-IL

composite membranes with the best CO2 separation performance.

The data of PIL C(CN)3 – 60 IL C(CN)3 membrane was taken from

Tomé et al. [20] The listed uncertainties represent the standard

deviations, based on three experiments.

Composite Membranes PCO2 (Barrer) αCO2/N2

PIL N(CN)2 – 60 IL C(CN)3 249.0 ± 1.0 61.3 ± 0.8

PIL C(CN)3 – 60 IL C(CN)3 439.3 ± 0.1 55.9 ± 0.1

PIL C(CN)3 – 60 IL B(CN)4 472.7 ± 3.0 54.4 ± 0.8

PIL B(CN)4 – 60 IL C(CN)3 502.1 ± 1.9 43.1± 0.4

The results provided in Table 2 display a trade-

off between the CO2 permeability and the CO2/N2

permselectivity values, in other words, when the CO2

permeability increases the permselectivity value

decreases. The change in CO2/N2 permselectivity can

be attributed to a solubility controlled mechanism

since the diffusivity selectivity (D CO2/N2) values fall

in the 0.5 – 0.7 range, while the solubility selectivity

(S CO2/N2) values range from 73.0 to 100.2. The same

behavior was also observed for the previously reported

composite membranes bearing PIL C(CN)3 with 20,

40 and 60 wt% of the free IL C(CN)3 composite

membranes. [20]

9

In sum, it can be concluded that the increase of

CO2 permeability is generally controlled by gas

diffusivity, while the increase in CO2/N2

permselectivity is due to contributions of a solubility

controlled mechanism.

All the results obtained for the four composite

membranes shown in Table 2 fall right above or on top

the upper-bound (Figure 8). Interestingly, the distance

between the membranes PIL N(CN)2 – 60 IL C(CN)3,

PIL C(CN)3 – 60 IL B(CN)4 and PIL B(CN)4 – 60 IL

C(CN)3 somewhat mimics the distance between the IL

N(CN)2, IL C(CN)3 and IL B(CN)4. In other words,

each CO2 separation performance of the composite

membranes falls back to slightly lower permeability

and permselectivity values than that of the respective

SILMs, due to the presence of polymeric PIL chains in

the membrane (Figure 8, the fall backs are represented

by arrows). Nevertheless, these composite membranes

have very good CO2 separation performances, in

addition to their robustness advantages when

compared to SILMs. With further research on their

chemical and physical properties these PIL-IL

composite membranes may hold a great promise for

gas membrane technology, in particular for CO2/N2

separation.

Figure 8 - CO2 separation performance of the best PIL-IL

composite membranes studied in this work, plotted on a CO2/N2

Robeson plot. “Literature” stand for several neat PIL and PIL-IL

composite membranes previously reported by other research

groups. [19, 20, 22, 24, 29, 30-38] The data for the membrane PIL

C(CN)3 – 60 IL C(CN)3 was taken from reference [20] while the

ILs data came from references [26, 28]. The different data are

plotted on a log-log scale and the upper bound is adapted from

Robeson [10]. It should be noted that the two “Literature”

membranes (gray triangles) above the upper bound were measured

at low CO2 partial pressure and 95% RH.

4. Conclusions

The main goal of this work was the evaluation of

the CO2/N2 separation performance of several PIL-IL

composite membranes.

All membranes here prepared have cyano-

functionalized counter anions ([N(CN)2]-, [C(CN)3]- or

[B(CN)4]-) or an [NTf2]- counter anion, which are

different in the PIL and IL. For this purpose, four PILs

having a pyrrolidinium polycation were synthesized

via simple and straightforward anion exchange

reactions. The four PILs were then used to prepare

PIL-IL composite membranes, by blending the PILs

with different weight percentages of five

commercially available free ILs, and then applying a

solvent casting process From all 42 tested membranes,

21 were free standing and homogenous. Membranes

with the [C(CN)3]- anion (either in the PIL or IL) had

a membrane formation success rate of 71%, while

membranes with the [N(CN)2]- anion had the lowest

formation success rate, 33%. The versatility,

flexibility and chemical compatibility of [C(CN)3]-

anion may be the reason why most membranes that

contain this anion are homogenous and free standing.

For all membranes tested, the CO2 permeability was

always higher than the N2 permeability thanks to a

correspondingly difference between these two gases

solubilities. When increasing the free IL content of the

membranes, both CO2 and N2 permeabilities also

increased. This increment in gas permeability was

attributed to an increase in gas diffusivity. It was also

observed that, while the increase/decrease of CO2 and

N2 permeability is a diffusivity controlled mechanism,

the increase/decrease of CO2/N2 permselectivity is a

solubility controlled mechanism. The three composite

membranes PIL N(CN)2 – 60 IL C(CN)3, PIL C(CN)3

– 60 IL B(CN)4 and PIL B(CN)4 – 60 IL C(CN)3, were

on top or even surpassed the Robeson 2008 upper

bound for CO2/N2 separation. It was found that

replacing the [C(CN)3]- anion for the [B(CN)4]- anion

increases gas permeability. Further research of the

composite membranes with the best CO2 separation

performance under real industrial conditions should

also be considered. The evaluation of the behavior of

these membranes under different pressure,

temperature, compositions of binary gas mixtures

(including the exposure to different impurities) and

humidity contents is crucial for their application in

CO2 separation processes.

References

[1] S. D. Kenarsari, D. Yang, G. Jiang, S. Zhang, J.

Wang, A. G. Russell, Q. Wei and M. Fan, RSC

Adv., 2013.

[2] M. E. Boot-Handford, J. C. Abanades, E. J.

Anthony, M. J. Blunt, S. Brandani, N. Mac Dowell,

J. R. Fernandez, M.-C. Ferrari, R. Gross, J. P.

Hallett, R. S. Haszeldine, P. Heptonstall, A.

Lyngfelt, Z. Makuch, E. Mangano, R. T. J. Porter,

M. Pourkashanian, G. T. Rochelle, N. Shah, J. G.

Yao and P. S. Fennell, Energ. Environ. Sci., 2014,

7, 130-189.

[3] http://www.ccsassociation.org/ (15 April 2016)

1

10

100

1000

10 100 1000

αC

O2/N

2

P CO2 (Barrer)

Upper Bound (Robeson 2008) Literature

PIL N(CN)2 - 60 IL C(CN)3 PIL C(CN)3 - 60 IL C(CN)3

PIL C(CN)3 - 60 IL B(CN)4 PIL B(CN)4 - 60 IL C(CN)3

IL N(CN)2 IL C(CN)3

IL B(CN)4

10

[4] N. MacDowell, N. Florin, A. Buchard, J. Hallett,

A. Galindo, G. Jackson, C. S. Adjiman, C. K.

Williams, N. Shah and P. Fennell, Energ. Environ.

Sci., 2010, 3, 1645-1669.

[5] Brunetti, F. Scura, G. Barbieri and E. Drioli, J.

Membr. Sci., 2010, 359(1), 115-125.

[6] R. W. Baker, Membrane Technology and

Applications, 2nd ed., John Wiley & Sons Ltd.,

Chischester, England, 2004.

[7] D. F. Sanders, Z. P. Smith, R. Guo, L. M. Robeson,

J. E. McGrath, D. R. Paul and B. D. Freeman,

Polymer, 2013, 54, 4729-4761.

[8] B. D. Freeman, Macromolecules, 1999, 32, 375-

380.

[9] L. M. Robeson, J. Membr. Sci., 1991, 62, 165-185.

[10] L. M. Robeson, J. Membr. Sci., 2008, 320, 390-

400.

[11] N. V. Plechkova and K. R. Seddon, Chem. Soc.

Rev., 2008, 37, 123-150.

[12] L. C. Tomé and I. M. Marrucho, Chem. Soc. Rev.,

2016, 45, 2785-2824.

[13] S. Zulfiqar, M. I. Sarwar and D. Mecerreyes,

Polym. Chem., 2015, 6, 6435–6451

[14] J. Yuan, D. Mecerreyes and M. Antonietti,

Progress in Polymer Science, 2013, 38.7, 1009-

1036.

[15] T. D. Ho, A. J. Canestraro and J. L. Anderson,

Anal. Chim. Acta, 2011, 695, 18-43.

[16] S. T. Hemp, M. H. Allen, M. D. Green and T. E.

Long, Biomacromolecules, 2012, 13, 231-238.

[17] Y. Kohno and H. Ohno, Aust. J. Chem., 2012, 65,

91-94.

[18] L. C. Tomé, D. Mecerreyes, C. S. R. Freire, L. P.

N. Rebelo and I. M. Marrucho, J. Mater. Chem. A,

2014, 2, 5631-5639.

[19] L. C. Tomé, D. Mecerreyes, C. S. R. Freire, L. P.

N. Rebelo, I. M. Marrucho, J. Membr. Sci., 2013,

428, 260–266.

[20] L. C. Tome, M. Isik, C. S. R. Freire, D. Mecerreyes

and I. M. Marrucho, J. Membr. Sci., 2015, 483,

155-165.

[21] R. D. Noble, D. L. Gin, J. Membr. Sci., 2011, 369,

1–4.

[22] J. E. Bara, E. S. Hatakeyama, D. L. Gin and R. D.

Noble, Polym. Adv. Technol., 2008, 19, 1415-1420.

[23] L. C. Tomé, A. S. L. Gouveia, C. S. R. Freire, D.

Mecerreyes and I. M. Marrucho, J. Membr. Sci.,

2015, 486, 40-48.

[24] P. Li, K. P. Pramoda and T.-S. Chung, Ind. Eng.

Chem. Res., 2011, 50, 9344-9353.

[25] H. Liu, S. Dai and D. Jiang,

Phys.Chem.Chem.Phys., 2014, 16, 1909-1913.

[26] L. C. Tomé, C. Florindo, C. S. R. Freire, L. P. N.

Rebelo and I. M. Marrucho, Phys. Chem. Chem.

Phys., 2014, 16, 17172-17182.

[27] S. M. Mahurin, J. S. Lee, G. A. Baker, H. Luo and

S. Dai, J. Membr. Sci., 2010, 353, 177-183.

[28] H. Tokuda, S. Tsuzuki, M. A. B. H Susas, K.

Hayamizu, and M. Watanabe, J. Phys. Chem.

B, 2006, 110(39), 19593-19600.

[29] J. E. Bara, S. Lessmann, C. J. Gabriel, E. S.

Hatakeyama, R. D. Noble and D. L. Gin, Ind. Eng.

Chem. Res., 2007, 46, 5397-5404.

[30] J. E. Bara, C. J. Gabriel, E. S. Hatakeyama, T. K.

Carlisle, S. Lessmann, R. D. Noble and D. L. Gin,

J. Membr. Sci., 2008, 321, 3-7.

[31] J. E. Bara, E. S. Hatakeyama, C. J. Gabriel, X.

Zeng, S. Lessmann, D. L. Gin and R. D. Noble, J.

Membr. Sci., 2008, 316, 186-191.

[32] T. K. Carlisle, E. F. Wiesenauer, G. D. Nicodemus,

D. L. Gin and R. D. Noble, Ind. Eng. Chem. Res.,

2013, 52, 1023-1032.

[33] J. E. Bara, D. L. Gin and R. D. Noble, Ind. Eng.

Chem. Res., 2008, 47, 9919-9924.

[34] J. E. Bara, R. D. Noble and D. L. Gin, Ind. Eng.

Chem. Res., 2009, 48, 4607-4610.

[35] P. Li, D. R. Paul and T. S. Chung., Green Chem.,

2012, 14, 1052-1063.

[36] T. K. Carlisle, J. E. Bara, A. L. Lafrate, D. L. Gin,

R. D. Noble, J. Membr. Sci., 2010, 359, 37-43.

[37] L. C. Tomé, M. A. Aboudzadeh, L. P. N. Rebelo, C.

S. R. Freire, D. Mecerreyes and I. M. Marrucho, J.

Mater. Chem. A, 2013, 1, 10403-10411.

[38] W. S. Chi, S. U. Hong, B. Jung, S. W. Kang, Y. S.

Kang and J. H. Kim, J. Membr. Sci., 2013, 443, 54-

61.

[39] L.A. Blanchard, Z. Y. Gu, J. F. Brennecke, J.

Physical Chem., 2001, B105(12), 2437–2444.